Andrew Saxe
Professor of Theoretical Neuroscience and Machine Learning
Develops the theory of deep learning to analyse learning dynamics in artificial networks and in the brain.
Andrew Saxe's lab at UCL works on computational neuroscience & cognition, learning theory & optimization and neuroscience & brain-computer interfaces. 34 papers since 2023, 23 at top venues, mostly ICML and NeurIPS. 9 PhD students. 10 alumni.
At a glance
- papers, 2023–2026
- 9/ year
- at T1 venues
- 68%
- PhD students
- 9
- MS students
- 0
- alumni listed
- 10
Papers by yearfrom the lab homepage, 2023–2026
- 4 papers2026 · T1 3 · collecting
- 13 papers2025 · T1 9
- 9 papers2024 · T1 8
- 8 papers2023 · T1 3
T1 venuesOtherVenues: ICML, NeurIPS, ICLR, Neuron, Cell
Research focuspaper titles that use each topic's words
- Learning theory & optimization20 papers
- Computational neuroscience & cognition1 paper
- Neuroscience & brain-computer interfaces0 papers
- gatsby0 papers
- gatsby computational0 papers
Paper list
From the lab homepage.
20264 papers
- Proceedings of the Analytical Connectionism SchoolsAn Introduction to Connectionist Theories of Semantic Cognition (opens in a new tab)
- ICLR (T1)Comparing the learning dynamics of in-context learning and fine-tuning in language models (opens in a new tab)
- PNAS (T1)Compositionality and systematicity emerge from iterated learning in deep linear networks (opens in a new tab)
- ICLR (T1)Saddle-to-Saddle Dynamics Explains A Simplicity Bias Across Neural Network Architectures (opens in a new tab)
202513 papers
- Journal of Statistical Mechanics: Theory and ExperimentA theory of initialisation’s impact on specialisation (opens in a new tab)
- ICML (T1)Algorithm Development in Neural Networks: Insights from the Streaming Parity Task (opens in a new tab)
- Cell (T1)Dopamine encodes deep network teaching signals for individual learning trajectories (opens in a new tab)
- CCN : Annual Conference on Cognitive Computational NeuroscienceDynamics and structure of generalization during reinforcement learning in human brains and artificial networks (opens in a new tab)
- ICLR (T1)From lazy to rich: Exact learning dynamics in deep linear networks (opens in a new tab)
- ICLR (T1)Make Haste Slowly: A Theory of Emergent Structured Mixed Selectivity in Feature Learning ReLU Networks (opens in a new tab)
- NeurIPS (T1)Memory by accident: a theory of learning as a byproduct of network stabilization (opens in a new tab)
- ICML (T1)Not all solutions are created equal: An analytical dissociation of functional and representational similarity in deep linear neural networks (opens in a new tab)
- Phys. Rev. XRL Perceptron: Generalization Dynamics of Policy Learning in High Dimensions (opens in a new tab)
- CCN : Annual Conference on Cognitive Computational NeuroscienceStage-like Emergence of Task Strategies in Animals and in Neural Networks Trained by Gradient Descent (opens in a new tab)
- ICML (T1)Strategy Coopetition Explains the Emergence and Transience of In-Context Learning (opens in a new tab)
- ICML (T1)Training Dynamics of In-Context Learning in Linear Attention (opens in a new tab)
20249 papers
- PLOS Computational BiologyAbrupt and spontaneous strategy switches emerge in simple regularised neural networks (opens in a new tab)
- NeurIPS (T1)Flexible task abstractions emerge in linear networks with fast and bounded units (opens in a new tab)
- NeurIPS (T1)Get rich quick: exact solutions reveal how unbalanced initializations promote rapid feature learning (opens in a new tab)
- NeurIPS (T1)Nonlinear dynamics of localization in neural receptive fields (opens in a new tab)
- ICML (T1)Tilting the Odds at the Lottery: the Interplay of Overparameterisation and Curricula in Neural Networks (opens in a new tab)
- ICML (T1)Understanding Unimodal Bias in Multimodal Deep Linear Networks (opens in a new tab)
- ICML (T1)What needs to go right for an induction head? A mechanistic study of in-context learning circuits and their formation (opens in a new tab)
- ICML (T1)When Representations Align: Universality in Representation Learning Dynamics (opens in a new tab)
- ICML (T1)Why Do Animals Need Shaping? A Theory of Task Composition and Curriculum Learning (opens in a new tab)
20238 papers
- Trends in NeurosciencesContinual task learning in natural and artificial agents (opens in a new tab)
- EACLKnow your audience: specializing grounded language models with listener subtraction (opens in a new tab)
- Neuron (T1)Mice identify subgoal locations through an action-driven mapping process (opens in a new tab)
- PLoS Computational BiologyModelling continual learning in humans with Hebbian context gating and exponentially decaying task signals (opens in a new tab)
- Neuron (T1)Neural knowledge assembly in humans and neural networks (opens in a new tab)
- Nature NeuroscienceOrganizing memories for generalization in complementary learning systems (opens in a new tab)
- ELifeStrategically managing learning during perceptual decision making (opens in a new tab)
- NeurIPS (T1)The Transient Nature of Emergent In-Context Learning in Transformers (opens in a new tab)
Where alumni went10 listed
Apply to UCLFall 2027
Not announced yet
Computer Science MPhil/PhD
- Round
- Main
- Opens
- —
- GRE
- See program page
- English
- See program page
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- USNYU
Jennifer Hill
- Probabilistic & causal inference
- USUCSD
Biwei Huang
- Probabilistic & causal inference
- SGNUS
Bryan Kian Hsiang Low
- Probabilistic & causal inference
- USUCSD
Yian Ma
- Reinforcement learning
- CNTsinghua
Hongning Wang
- Reinforcement learning
- SGNTU
Yingzhen Li
- Probabilistic & causal inference
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.